CN102512186B - Modeling of pharmaceutical propagation and response of patients to medication injection - Google Patents

Abstract

The invention refers to determining patient transfer functions ad response to of modeling patient to drug injection. A method of delivering a contrast enhancing fluid to a patient using an injector system, including: determining at least one patient transfer function for the patient based upon data specific to the patient, the at least one patient transfer function providing a time enhancement output for a given input and the data specific to the patient includes estimation of patient physiological parameters, estimation of patient anatomy parameter and at least one enhancement caused by injection of radiography enhancement liquid to a patient; (b) determining a desired time enhancement output; using the at least one patient transfer function to determine an injection procedure input; and controlling the injector system at least in part on the basis of the determined injection procedure input.

Description

Determine that patient's transfer function and modeling patient are to the response of drug injection

Patent application of the present invention is international filing date is on November 16th, 2005, application number is 200580038906.1, be entitled as the divisional application of the Chinese invention patent application of " determine patient's transfer function and respond to the patient of drug injection the system and method carrying out modeling ".

The cross reference of related application

This application claims the priority of the U.S. Provisional Patent Application of the serial number 60/628,201 that on November 16th, 2004 submits to, the disclosure of this application is incorporated herein by reference.

Background technology

The present invention relates generally to the modeling of the propagation of medicine in patient body, particularly relate to the modeling of the propagation of contrast agent in patient body for imaging process.

The list of references enumerated herein contributes to understanding the present invention or background of the present invention.But, comprise list of references herein and not intended to be permits that this list of references can be used as prior art relative to the present invention, and do not form this license.

Various contrast agent injectable enters in patient body, for various Diagnosis and Treat imaging process, such as X-ray procedure (comprising such as angiography, venography and urography), computer tomography (CT), nuclear magnetic resonance (MRI), ultra sonic imaging, based on the imaging of light and positron emission computerized tomography (PET).Such as, CT scanner has become indispensable modern diagnosis imaging tool.It makes to carry out accurate measurement of 2,3 and 4 dimensions to the physiological process in anatomical structure and some situations becomes possibility.The not easy-to-use CT scanner of the imaging of soft tissue, vascular system and other structure is realized, because X-ray can not differentially be decayed to suitable degree by these structures.In order to overcome these restrictions, usually absorption ray medicine or radiopaque medicine or contrast agent are injected in peripheral venous circulation.The atomic iodine of three or more is normally bonded in the water soluble salt in benzene structure by the contrast agent for CT imaging.X-ray in the energy range used in iodine decay medical imaging process.Before scanning, computer-controlled pump or syringe by the contrast agent of precise volume to be usually expelled to the Venous system of patient from the flow velocity of 0.5 to 6ml/s (pressure of generation is up to 300psi).The example being generally used for front year aspirating syringe in CT process is such as awarding the U.S. Patent No. 5,300,031,5,383,858 and 6,652 of giving assignee of the present invention, and open in 489, these patents are incorporated herein by reference.

Now, multi-detector CT scanner (MDCT) allows clinician to carry out non-parallel anatomy and physiological diagnostic scan to patient.But use this new technique, new challenge is created again to daily use.Although cover at volume and image resolution ratio has breakthrough, CT scanner of new generation still needs to realize optimized image and diagnosis with Iodine contrast medium.In addition, scanning is timed to the importance consistent with the best contrast concentration moment to increase in MDCT situation.

Medicine is not said in conjunction with the meaning its control program with physiological interactional understanding or estimation from injecting systems, the conveying of contrast agent normally open loop.Injecting systems accurately carries the contrast agent of the amount being programmed for concrete speed.This method scan carry out the quite a while make early stage drug metabolism not affect the quality of diagnostic scan time work good.The method is to perfusion in the target of scanning, good by also working during estimation in infusion of medicine such as parenchyma or suspicious cancer.The progress of scanning technique makes image can obtain (second) within the extremely short period.Need to consider early stage drug metabolism dynamic characteristic and the drug effect dynamic characteristic of contrast agent together with this trend presents with the volume more and more expecting to produce anatomical structure (similar heart, its coronary with lead to heart or the large vascular from heart).Ideally, because occurring in trunk that attenuation curve that contrast agent produces preferably on the multiple position of patient balanced (smooth) fully similar, so that volume presents and Accurate Diagnosis, and it is the image scanning carried out in timing best contrast concentration moment in region of interest.

In imaging and other process, the difference of the dose requirements of different patient is realized.Such as, award the U.S. Patent No. 5,840,026 of giving assignee of the present invention and disclose the concrete data of patient obtained before use is injected or during injection customize injection apparatus and method to patient.Although be realized based on the dose requirements difference of the medical imaging process of patient's difference, general medical imaging process still continues to use predetermined close or the standard transmission protocol for injection of contrast medium in medical imaging process.Consider the recent available CT scanner comprising MDCT scanner scanning speed increase, single-phase be injected at uses in the field of this rapid scanning than two-phase inject be dominant.Although use fixing host-host protocol (be no matter single-phase, two-phase or heterogeneous) to simplify process, under identical agreement, provide the contrast agent of same dose can produce the picture contrast result all extremely different with quality to different patient.In addition, by introducing up-to-date MDCT scanner, an open question in clinical practice and in CT document whether can divert from one use to another process in using MDCT machine well for the standard contrast agent agreement of monolithic, helical scanner.See " Sixteen-row multislice computed tomography:basic concepts; protocols, and enhanced clinical applications (16 row multi-disc computed tomography visualizations: basic conception, agreement and enhanced medical applications) " that the people such as F.Cademartiri and G.Luccichenti deliver for 2004 on 25 (1) the volume 2-16 pages of Semin Ultrasound CT MR.

Some researchs have been attempted carrying out quantitative analysis to injection process in CT angiography (CTA) process, to improve and to estimate that tremulous pulse strengthens.Such as, Bae and colleague thereof develop pharmacokinetic (PK) and the dynamic model of contrast agent behavior, and have solved coupled differential system to find the driving function causing the most even tremulous pulse to strengthen.K.T.Bae, " Aortic and hepatic contrast medium enhancement at CT.Part I.Prediction with a computer model (carrying out the enhancing of the pre-timing aorta of CT part i and liver contrast agent with computer model) " of J.P.Heiken and J.A.Brink, Radiology, 207 volume 647-55 pages, 1998, K.T.Bae " Peak contrast enhancement in CT and MR angiography:when does it occur and why? Pharmacokinetic study in a porcine model (does contrast peak in CT and MR angiography strengthen: when occur and why? take pig as the drug metabolism Research on Dynamic Characteristic of model) ", Radiology, 227 volume 809-16 pages, 2003, " Multiphasic Injection Method for Uniform Prolonged Vascular Enhancement at CT Angiography:Pharmacokinetic Analysis and Experimental Porcine Method (for evenly extending the heterogeneous injection that vascular strengthens in the CT angiography: pig method is used in drug metabolism dynamic analysis and test) " of K.T.Bae etc., Radiology, 216 volume 872-880 pages, 2000, U.S. Patent No. 5, 583, 902, 5, 687, 208, 6, 055, 985, 6, 470, 889 and 6, 635, 030, their disclosure is incorporated herein by reference.The inverse solution of the differential equation group of the simplification compartment model of being set forth by the people such as Bae represents: the contrast agent flow velocity that index reduces can cause the best in CT imaging process/fixing to strengthen.

The PK method strengthened for deriving from homogeneous image of Bae depend on for clinician not easily can many physiological parameters, such as CBV, diffusion rate and cardiac output.Not there is the significant drawback that kinemic clear and definite measured value is Bae method, although attempt carrying out this value approximate based on patient age, body weight and height.In addition, the realization of PK model in controller frame is not considered.Injection distribution (profile) calculated by the anti-solution of PK model is the distribution that CT motoring syringe institute is not easy realization when not carrying out main change.In addition, pulsatile flow, vascular flexibility and local blood/contrast agent parameter (i.e. viscosity) do not considered by the PK model of Bae.

Cardiovascular physiology and contrast agent dynamics characteristic are considered as " flight data recorder " by Fleischmann and colleague thereof, and by forcing this system to determine its impulse response (impulse response) with the contrast agent (approximate unit pulse) of a bolus.In the method, paired pulses response is carried out Fourier transform and is controlled the estimation of this transforming function transformation function to find optimal injection track." Mathematical analysis of arterial enhancement and optimization of bolus geometry for CT angiography using the discrete Fourier transform (the using discrete Fourier transform strengthen the tremulous pulse of CT angiography and inject the optimized mathematical analysis of agglomerate geometry) " of D.Fleischmann and K.Hittmair, J.ComputAssist Tomogr, 23 volume 474-84 pages, 1999, its disclosure was incorporated herein by reference.

The usage of contrast agent is normally single-phase-and be the contrast agent of 100 to 150mL under a flow velocity, this causes uneven enhancing curve.For example, referring to the document of above-mentioned D.Fleischmann and K.Hittmair; And K.T.Bae " Peak contrast enhancement in CT and MR angiography:when does it occur and why? Pharmacokinetic study in a porcine model (does the peak value radiography in CT and MR angiography strengthen: when occur and why occur? the drug metabolism Research on Dynamic Characteristic in pig model) ", Radiology, 227 volume 809-16 pages, 2003, these documents were incorporated herein by reference.Fleischmann and Hitmmair proposes a kind of scheme that be intended to the imaging of optimization large artery trunks, that attempt the usage of contrast agent to be adjusted to the two-phase injection being suitable for single patient.The basic difficulty that control CT contrast agent occurs is that hypertonicity medicine is soon from the diffusion of central blood room.In addition, contrast agent mixes with the blood not containing contrast agent and is diluted.The mixing of contrast agent and dilution are by such as the illustrative peak value enhancing of Fig. 1 and distortion strengthen curve reflection.

Fleischmann forbids that (test injection) (16ml contrast agent under 4ml/s) of once injecting on a small quantity of contrast agent carried out before diagnostic scan.Dynamic contrast enhancement is carried out on interested vascular.The treated scan-data (test scan) of gained is construed as the impulse response of patient/contrast agent system.Fleischmann is by obtaining the Fourier transform of patient's transfer function divided by the Fourier transform of test injection by the Fourier transform of test scan.Suppose that this system is linear time invariant (LTI) system and desired output time-domain signal is known (the smooth diagnostic scan under predetermined enhanced level), Fleischmann is by obtaining inputted time signal by the frequency domain representation of desired output divided by the frequency domain representation of patient's transfer function.

The method of Fleischmann wishes to be that it obtains the fact of the expression of patient based on known test injection.Because the method for the people such as Fleischmann calculates the in fact irrealizable input signal due to injecting systems restriction (such as, flow velocity limits), so must carry out truncate to calculated continuous time signal and be similar to.Because of the inaccuracy introduced by this step, so the idealized input track calculated is not best.In addition, do not know whether linear hypothesis is set up all patients and pathology.Finally, also whether the unclear enhancing curve produced by his method is more even than injecting by simple two-phase the enhancing curve produced.

For the medicine except contrast agent, also have been developed different models.Such as, " Optimal insulin infusion resulting from a mathematical model of blood glucose dynamics (the best infusion of insulin that the mathematical model because of blood glucose dynamic characteristic causes) " of Fisher and Teo, IEEE Trans Biomed Eng, 36 (4) volume 479-486 pages, 1989, the document is incorporated herein by reference, its simulation glucose and insulin dynamic characteristic, be intended to produce best infusion of insulin parameter.They carry (Ricatti) equation by applying quadratic form performance standard and solving algebraically multitude card and this issue handling are become classic optimisation problem.They find that comparing Pulse Width Control with the CI of injection of insulin, sub optimal control with without adjustment is the best approach.

" Algorithm for optimal linear model-based control with application to pharmacokinetic model-driven drug delivery (based on the control algolithm of optimum linearity model and the application in the drug conveying of drug metabolism dynamic performance model driving thereof) " of Jacobs, IEEE Trans Biomed Eng, 37 (1) volume 107-109 pages, nineteen ninety, propose and be provided with drug model concurrently with actual drug process, for the control algolithm that anaesthetic regulates, the document is incorporated herein by reference.Clinician determines target plasma concentration.

" The hybrid model:a new pharmacokinetic model for computer-controlled infusion pumps (mixed model: the novel drugs metabolism dynamic performance model controlling infusion pump for computer) " of Wada and Ward, IEEE Trans.Biomed Eng, 41 (2) volume 134-142 pages, 1994, be derived the 3 Room drug metabolism dynamic performance model that the method that adopts with Bee is similar, and use it in mixture control scheme to attempt regulating the plasma concentration (loaded with diluted upload alienating) of anaesthetic, the document is incorporated herein by reference.They attempt simulating the recirculation effect of contrast agent by blood flow, and wherein they simulate blood flow by transmission delay being inserted in its simulation.They can produce the simulation that forecast error is less than 5%.

" Open loop control of multiple drug effects in anesthesia the opened loop control of multi-medicament effect (during the anesthesia) " of Wada and Ward, IEEE Trans.Biomed Eng, 42 (7) volume 666-677 pages, nineteen ninety-five, also by its hybrid medicine metabolism dynamic characteristic (PK) model for controlling the multiple effect of anaesthetic, the document is incorporated herein by reference.Their control program needs anaesthetist to set the side effect level (being expressed as plasma concentration) of permission.

" Estimation of plasma insulin from plasma glucose (according to the plasma glucose estimation plasma insulin) " of Neatpisarnvanit and Boston, IEEE Trans Biomed Eng, 49 (11) volume 1253-1259 pages, 2002, application recurrence least square number estimation algorithm predicts the plasma concentration of glucose and insulin, and the document is incorporated herein by reference.Their method produces the prediction having the blood plasma level of 6 intraindividual glucoses and insulin to mate (experimental data holds resistance to test by intravenous glucose) in 7 patients, and advantageously conform to." A new paradigm for the closed-loop intraoperative administration of analgesics in humans (analgesic in operation to the new example of the closed loop medication of human body) " of the people such as Gentilini, IEEE Tran Biomed Eng, 49 (4) volume 289-299 pages, 2002, propose Model Predictive Control (MPC) method of the plasma concentration for being controlled anesthetics alfentanil (alfentanil) by computer-controlled infusion pump, the document is incorporated herein by reference.This drug metabolism dynamic performance model is the 3-room model describing anaesthetic distribution in human body.Controller depends on the observer of the plasma concentration according to the measured value of average artery pressure and the PK model assessment medicine of parallel running.The people such as Gentilini are excessive to prevent to Cmax setting restriction.Their disturbance also in filtering average artery pressure measured value and allow controller to depend on the state (namely hypotension is to hypertension) of patient is run sooner or more slowly.

Summary of the invention

The invention provides and be convenient to determine/set up or regulate patient's transfer function or patient to general modifying device, the system and method for the model (or model parameter) of the response of drug injection.Patient's transfer function or model can such as based on or the information measured known from the feedback test injection and/or this process before the procedure starts, to improve or to optimize medicine transmission (contrast concentration such as, in one or more region of interest).

In an aspect, the invention provides and use injector system radiography to be strengthened the method that liquid flows to patient, comprise: at least one the patient's transfer function determining this patient based on the concrete data of patient, this at least one patient's transfer function provides the time to strengthen to given input and exports; Determine that the expected time strengthens to export; Use this at least one patient's transfer function to determine that injection process inputs; And at least in part based on determined injection process input control injector system.Consider at least one operation restriction or constraint of injector system, can determine that injection process inputs.

At least one patient's transfer function is determined by such as using the system identification model comprising the parameter relevant to physiological parameters of patients.This system identification model preferably can discretization.

The method also can comprise the following steps: use the estimated value of at least one physiological parameter of patient to form initial patient's transfer function; Inject; And strengthen output correction patient transfer function based at least one time of this injection.At least one physiological parameters of patients can strengthen to export according at least one time to be measured.This injection can be the test injection of carrying out before diagnosing image process or the injection carried out in imaging process.

The time produced because of test injection strengthens and exports and can measure at least two different region of interest.Time strengthens at least one difference between exporting can such as provide measuring of at least one physiological parameters of patients.At least one physiological parameters of patients can be the parameter of cardiorespiratory system.This at least one physiological parameters of patients can be blood volume in such as cardiac output, a position, speed rates item or by postponing.In one embodiment, the very first time strengthens output and can measure in ascending aorta or descending aorta, and the enhancing output of the second time can be measured in pulmonary trunk.

This at least one patient's transfer function is also determined by following steps: collect correspond to because of liquid infusion produce time response curve data; And determine at least one mathematical model describing these data.

In one embodiment, this mathematical model is not deconvoluted by the continuous of these data or discrete time Fourier and determines.This model can be a parameter model.This model can be such as rolling average or automatic returning rolling average.This mathematical model can suppose linear and the time is constant.

This model also can be the nonparametric model determined by spectrum estimating techniques.These spectrum estimating techniques can be such as Welch method, Bartlett method, Multiple Signal Classification (MUSIC) method or periodogram (Periodogram) method.Data can be collected during the test injection at least one times before imaging injection.

The Data Update that at least one patient's transfer function of the present invention is collected during can be used on imaging injection.

As mentioned above, at least one patient's transfer function can be determined based on injecting at least one times at least in part.This is injected at least one times can be the test injection of carrying out before diagnosing image process.In one embodiment, this test injection comprises contrast-medium injection and the injection of non-contrast afterwards.Non-contrast can be injected with at the volume flow rate injecting the contrast agent flow velocity before non-contrast substantially identical.Non-contrast can be saline.

More than one test injection can be carried out.Such as, test injection can only include contrast-medium injection and another test injection can comprise contrast-medium injection and the injection of non-contrast afterwards.

Injection process input of the present invention is by using analytic solutions or using numerical value constraint optimization to determine.In one embodiment, numerical value constraint optimization is weighted least-squares numerical optimization.

Injection process input can such as be relevant to one or more consideration and be optimized.Such as, the quality that injection process input can be optimized to contrast medium in the contrast agent enhanced liquid making to be transported to patient minimizes.

The example being suitable for the contrast medium used together in conjunction with the present invention includes but not limited to iodine, hernia and gadolinium.It can be that such as CT radiography strengthens liquid, MRI radiography strengthens liquid, ultrasonic Enhanced Imaging liquid or radioactivity radiography and strengthens liquid that radiography strengthens liquid.

In an embodiment of the present invention, at least two patient's transfer functions are determined and injection process input is determined based on one of patient's transfer function.Such as, first patient's transfer function can use the system identification model comprising the parameter relevant with physiological parameters of patients to determine, and second patient's transfer function can use and strengthens the mathematical model that the corresponding data of curve determine by collecting determine with because injecting the time produced, this mathematical model describes this data.Can to such as which patient's transfer function provide given input to export to result between relevant the making of the best determine.

On the other hand, the invention provides the method determining at least one physiological parameters of patients from imaging process, comprise: at least two different region of interest Measuring Time are strengthened to exporting, and the time of determining strengthens at least one difference between exporting to provide measuring of at least one physiological parameters of patients.This at least one physiological parameters of patients can be the parameter of such as cardiorespiratory system.This at least one physiological parameters of patients can be blood volume in such as cardiac output, a position, speed rates item or by postponing.In one embodiment, the very first time strengthens to export measures in ascending aorta or descending aorta, and the enhancing output of the second time is measured in pulmonary trunk.

More on the one hand in, the invention provides for by Liquid transfer to the injector system of patient, comprising: syringe and the controller communicated to connect with syringe.At least one patient's transfer function that this controller comprises (such as, being stored in its memorizer be effectively connected) determines this patient according to the concrete data of patient.This at least one patient's transfer function provides the time to strengthen to given input and exports.This controller comprises processor (such as, digital microprocessor) and determines that injection process inputs to use at least one patient's transfer function to strengthen output to the expected time.

Injection process input is determined by least one physical restriction or constraint considering syringe.Injection process input can such as use analytic solutions or numerical value constraint optimization to determine.Numerical value constraint optimization can be such as weighted least-squares numerical optimization.Injection process input can be optimized, such as to minimize the quality that the radiography flowing to patient strengthens contrast medium in liquid.

Contrast medium can be such as iodine, xenon or gadolinium.It can be that such as CT radiography strengthens liquid, MRI radiography strengthens liquid, ultrasonic Enhanced Imaging liquid or radioactivity radiography and strengthens liquid that radiography strengthens liquid.

In another aspect, the invention provides an imaging system, comprising: imager, create the image of patient's region of interest; Syringe, is suitable for injection of contrast medium; And controller, communicate to connect to control this syringe with syringe.Controller comprises at least one the patient's transfer function determined this patient based on the concrete data of patient.This at least one patient's transfer function provides the time to strengthen to given input and exports.Controller also comprises processor, strengthens output determine that injection process inputs to use at least one patient's transfer function as above to the expected time.

In some embodiments of the present invention, the spectrum analysis of the patient's response/scan-data obtained during contrast agent is injected on a small quantity and parameter estimation are for developing the control example of the closed circulation control that can provide contrast agent usage.

In one aspect, the invention provides the modeling method that pharmaceutical liquid is propagated in patient body, comprising: collect the data corresponding with curve time response produced because of this liquid infusion; And determine at least one mathematical model describing these data.This mathematical model can be not such as by these data continuously or discrete time Fourier to deconvolute the model determined.

This model can be the parameter model of such as moving average model(MA model) or ARMA model.This model also can be the parameter model of the parameter fitting comprising measurement data.Once can such as in mathematical model hypothesis linear and the time is constant, then this model also can be the nonparametric model determined by spectrum estimating techniques.Suitable spectrum estimating techniques include but not limited to Welch method, Bartlett method, Multiple Signal Classification (MUSIC) method or period map method.

The liquid be injected can be the contrast agent used in such as imaging process, and the time that collected data can produce corresponding to the injection because of contrast agent strengthens curve.

Collected time response curve or the time data that strengthen curve can collect during at least one test injection before imaging injection.The data that this model is collected during also can be used on imaging (or other process) injection are determined and/or upgrade.In one embodiment, test injection comprises the injection of contrast agent and the injection of non-contrast subsequently.Non-contrast can be such as substantially the same with the contrast agent flow velocity before this non-contrast liquid infusion volume flow rate inject.This non-contrast can be such as saline.Once above test injection can be carried out.In such embodiment, test injection only includes the injection that another test injection of contrast-medium injection then comprises contrast-medium injection and non-contrast subsequently.

On the other hand, the invention provides and control to use syringe to the method for patient injection pharmaceutical liquid in medical procedure, comprising: collect the data corresponding with patient's response curve that the injection because of this liquid produces; Determine at least one mathematical model describing these data; And in medical procedure, control syringe to control the liquid infusion to patient, so that be at least partly based on Mathematical Models patient response.This mathematical model can be such as not by these data continuously or discrete time Fourier to deconvolute the model determined.

Medical procedure can be the medical imaging procedures such as using imagine scanner, and collected data may correspond to the time enhancing curve produced in the injection because of contrast agent.Controlled injection device to control contrast-medium injection to patient so that be at least partly based on the image of Mathematical Models region of interest.

Also syringe can be controlled based on the information about patient's response in imaging process at least in part.In addition, syringe can be controlled based on the information of at least one measurement physiologic variables of concerned patients at least in part.This measurement physiologic variables can be used for the output changing mathematical model.

In one embodiment, control the step of syringe to be included in a moment and to start injection of contrast medium and start the image scanning of region of interest in the second moment determined based on mathematical model at least in part.This second moment can be determined based on to the moment prediction arriving the predetermined enhanced level determined by mathematical model.

On the other hand, the invention provides an injecting systems, comprising: syringe; And injector controller, with syringe efficient communication to control syringe.This injector controller controls the injection of liquid based at least one mathematical model as above.Like this, mathematical model strengthens the corresponding data of curve with the time produced because of contrast-medium injection determine by collecting.This mathematical model can be such as not by these data continuously or discrete time Fourier to deconvolute the model determined.Controller can comprise the computer such as with at least one processing unit and at least one memorizer.Computer program in order to determine mathematical model stores wherein by memorizer.

More on the one hand, the invention provides and control to use syringe to the method for the contrast-medium injection of patient in the medical imaging process using imagine scanner, comprising: determine that the time that at least one mathematical model produces because of contrast-medium injection with prediction strengthens response; Determine that infusion protocol is to respond by determining that to mathematical model the constraint input solution scheduled time of coming in approximate patient body strengthens; And in medical imaging process, use infusion protocol to control syringe to control the contrast-medium injection to patient, thus set up the image of region of interest.

The method also can comprise the feedback owing to strengthening response about the time in imaging process and change the step of infusion protocol.The method also can comprise the step changing infusion protocol due to the data about at least one physiological parameters of patients in imaging process.

In one embodiment, determine that infusion protocol strengthens the step responded and realizes by using numerical value solver or numerical optimization device with the approximate scheduled time.At least one performance constraint that can be subject to such as syringe to the constraint input solution of mathematical model retrained.Be subject to or be selectively subject to retraining to patient safety or comfortable at least one relevant performance constraint to the constraint input Xie Yeke of mathematical model.

The injection of contrast agent can such as a moment, and the image scanning of region of interest can start in the second moment determined based on mathematical model at least in part.This second moment can such as be determined based on to the moment prediction reaching the predetermined enhanced level determined by mathematical model.

At least one mathematical model can be patient's transfer function of this patient based on the concrete data of patient.This patient's transfer function provides the time to strengthen to given input and exports.First patient's transfer function can such as strengthen by the time of collecting with produce because of patient injection the mathematics model of cognition that the corresponding data of curve determine determine by using the system identification model that comprises the parameter relevant to physiological parameters of patients or using, and wherein this mathematics model of cognition describes this data.On the other hand, the invention provides the system for realizing medical procedure, comprising: the sensing system detecting patient's response; Be suitable for the syringe of injectable drug liquid; And with this syringe efficient communication to control the controller of this syringe.This injector controller controls liquid infusion based at least one mathematical model.Mathematical model is determined by collecting the data corresponding with curve time response produced because of liquid infusion from sensing system.Mathematical model can be such as not by these data continuously or discrete time Fourier to deconvolute the model determined.

On the other hand, the invention provides an imaging system, comprising: the imager setting up the image of patient's region of interest; Be suitable for the syringe of injection of contrast medium; And with this syringe efficient communication to control the controller of this syringe.Injector controller controls the injection of contrast agent based at least one mathematical model.Mathematical model is determined by strengthening the corresponding data of curve from imager collection with the time produced because of contrast-medium injection.Mathematical model can be such as not by these data continuously or discrete time Fourier to deconvolute the model determined.

On the other hand, the invention provides the method controlling to use the syringe with the controller communicated to connect with computer storage to inject the pharmaceutical liquid of patient in medical procedures, comprising: collect the data corresponding with the patient's response curve produced because of liquid infusion; From the multiple mathematical modeies be stored in computer storage, select at least one mathematical model to describe this data; Adjust this model to be suitable for collected data; And in medical procedure, control syringe by controller, to control to respond down to being at least partly based on this Mathematical Models patient the liquid infusion of patient.

More on the one hand, the invention provides the system of the image for setting up patient's region of interest, comprising: for measuring the image device of patient characteristics on region of interest; For the syringe to patient injection medicine; Also at least one standard (or reference) position measured by image device; And based on standard location measured value adjustment or correct the computational algorithm of measured value of feature on interested patient position.Suitable standard location can outside patient body.Suitable standard location can be the position of patient.

In in another, the invention provides the method for the image for setting up region of interest in patient body, comprising: use image device to measure patient characteristics on region of interest; To patient injection medicine, measure at least one standard location with image device; And correct based on the measured value of standard location or adjust the measured value of feature on interested patient position.

The benefit provided by each embodiment of the present invention includes but not limited to: the enhancing for subsequent picture process is more consistent, reduce for some patient's contrast agent or liquid load, contrast agent dosage increases reaching enough picture contrast as required, ooze out reduce, image artifacts reduces, rephotography number of times reduces, all monolithics comprise optimized image contrast, observe the concordance increase in the scanning of disease or treatment progress in time and optionally imaging time is faster.

Accompanying drawing explanation

Other side of the present invention and its advantage are cognitive from the following detailed description when reading by reference to the accompanying drawings, in the accompanying drawings:

Figure 1A illustrates that the radiography Enhanced CT for blood vessel uses the typical time period of single-phase injection distribution acquisition to strengthen curve.

Figure 1B illustrates that the radiography Enhanced CT for blood vessel uses the typical time period of two-phase or two-phase injection distribution acquisition to strengthen curve.

Fig. 2 A represents the exemplary patient response to test injection.

Fig. 2 B illustrates the estimating impulse response of patient in time domain/contrast agent system, it by the discrete time Fourier that scanning device exports deconvolute divided by use derive from the radiography input function of the data of above-mentioned Fleischmann and Hittmar.

Fig. 2 C illustrates patient's impulse response, and wherein h (n) is from experience, and Dynamic CT data are from position interested in the descending aorta of people (Fleischmann and Hittmair, 1999 years).

Fig. 3 illustrates that the use contrast medium injector of MPC controller architecture of the present invention improves an embodiment of the enhancing of CT image.

Fig. 4 illustrates the reduction PK model of the X-ray contrast delivered in 1998 as people such as Bae, Heiken.

Fig. 5 illustrates the numerical solution of equation (2), and wherein the dynamic characteristic of 25 seconds is issued close to real recirculation phenomenon.

Fig. 6 illustrates that describing medicine is represented by the figure of physiology " mixing " model of cardiovascular system and lung capillaries bed diffusion transport.

Fig. 7 illustrates the model that the exponent number based on Fig. 6 model reduces, and wherein Qco represents cardiac output.

Fig. 8 illustrates the axial Dynamic CT image of pulmonary artery level, and wherein therefrom two region of interest (ROI) of extraction time enhancing curve are circled.

Fig. 9 illustrate saliferous and salt-free 20ml contrast agent bolus entered 239 pounds 64 years old women's body in after the Dynamic CT time strengthen curve, wherein the first curve illustrates the enhancement value in pulmonary trunk, and the second curve illustrates the enhancement value of ascending aorta.

Figure 10 illustrates the signal model for generation of the molecule in Prony method and denominator coefficients.

Figure 11 illustrates 64 point fast Fouriers conversion FFT (fs=.5Hz) taked h (n) in equation (1).

Figure 12 illustrate for spectrum estimation Welch method-by h (n) be used as input.

Figure 13 illustrates the estimation h for increasing exponent number _testn () and the Steiglitz-McBride being derived from system estimate the curve of the mean square deviation between the impulse response of (repeatedly 10 times).

Figure 14 illustrates the transfer function h for different rank _testthe Steiglitz-McBride estimation curve of (n).

Figure 15 A illustrates arm vein and leads to total figure of vein of heart.

Figure 15 B illustrates the some changes in arm vein anatomical structure.

Figure 16 A illustrates for the relative test enhancing in the pulmonary artery of the first patient.

Figure 16 B illustrates for the relative test enhancing in the ascending aorta of the first patient.

Figure 17 A illustrates for the relative test enhancing in the pulmonary artery of the second patient.

Figure 17 B illustrates for the relative test enhancing in the ascending aorta of the second patient.

Figure 18 A illustrates for the effective injection distribution of a patient without the test injection of normal saline washing.

Figure 18 B illustrates for the effective injection distribution of a different patient without the test injection of normal saline washing.

Figure 19 A illustrates the flow chart of conventional method of the present invention.

Figure 19 B illustrates the flow chart of an embodiment of the inventive method being combined with such as Fig. 7 model.

Detailed description of the invention

Fig. 1 illustrates that the typical time period using the single-phase radiography Enhanced CT of blood vessel to obtain strengthens curve.Unit HU is Hounsfield Units (Houndsfield Unit), namely changes into measuring of the X-radiation absorption intensity of signal intensity in image.Figure 1A illustrates and to strengthen at the peak value in about 45 second moment.In many imaging processes, the time strengthens curve preferably near specified level even (as shown in the thick black line in Figure 1A).When this curve uneven or smooth time, non-optimal image can cause error diagnostics in this imaging process.Because progressive permission of scanning technique obtains image within the shorter time, the uniformity strengthened on more long duration can reduce more or less in importance, but relative to contrast-medium injection to the suitable timing of scanning and to too much contrast agent or very few contrast agent avoid still important.

Figure 1B illustrates that the typical time period obtained by two-phase or the two-phase radiography Enhanced CT of blood vessel strengthens curve.This enhancing curve more or less more smooth or evenly.But the amount of flatness can change with patient, therefore still produce non-optimal image.

Fig. 2 A illustrates and responds the typical patient of test injection.Fig. 2 B and 2C illustrates typical patient's impulse response.Fig. 2 B is the response of time domain and Fig. 2 C is the h (n) in frequency domain disclosed in above-mentioned D.Fleischmann and K.Hittmair.Patient's impulse response is derived the response of test injection from patient by the Fourier de-convolution in such as frequency domain or time domain.In the injecting curve of Fig. 2 C, data within every 2 seconds, catch once by CT scanner.

In one embodiment of the present invention, Model Predictive Control (MPC) controller architecture strengthened for radiography has been set forth.In the embodiment shown in Fig. 3, control procedure uses the enhancing curve produced because of test injection such as to estimate when taking LTI system the parameter of multipole/zero model of patient system, if or in allowed time constant hypothesis, derive suitable kernel function.Also linear hypothesis can be relaxed easily.One simulation example of time varying system has its value relative to the resistor of time with the change of other independent variable of possibility and the circuit of capacitor.Usually, circuit analysis thinks when resistance and electric capacity are all fixed as single value.

Then, the parameter identified in test injection step can be used by PK/PD (medicine-dynamic characteristic) model, and this model is used for upgrading controller in full injecting step and target endpoint is such as predefined even enhancement value.As shown in Figure 3, controller also can receive and contribute to reducing the feedback signal from scanner (i.e. enhancement value (EV)) of controller error or the estimated parameter (i.e. heart rate (HR), blood pressure (BP), the breathing rate & degree of depth, patient body weight) from observer.Feedback from the control syringe of scanner is such as awarding the U.S. Patent No. 6,397 of giving assignee of the present invention, and describe in 098, the document is incorporated herein by reference.Model Predictive Control Algorithm such as can be embodied as the input trajectory regulating contrast agent to use based on the instantaneous enhancing data of collecting in a time step from scanner.If actual enhancement value is different (such as from the value predicted by the model produced in identification step, with regard to least square), then control algolithm scalable input flow velocity, to attempt making instantaneous enhancement value closer to the value by this model prediction in time step subsequently.Be aware of the patient heart rate obtained according to heart rate monitor (ECG), pulse oximeter or blood pressure monitor, the more high fidelity model of contrast agent propagation can be obtained.

In an embodiment of Fig. 3, model of the present invention produces estimation to patient's transfer function H (z) by zero pole point (pole-zero) modeling (ARMA technology), and row constraint numerical optimization of going forward side by side is to determine the input signal (i.e. infusion protocol) by producing desired output response (such as smooth enhanced ct scans-see Fig. 1).Or, RECURSIVE POLE PLACEMENT (pole-placement algorithm) can be used to control better to export response when the estimation of given H (z).

The structure of patient's transfer function is determined by analyzing patient's impulse response h (n) of collecting during clinical examination.Obviously, also not to the open analysis of the basic spectral content of patient's transfer function.ARMA modeling technique can such as generation of the coefficient of the reasonable transfer function of following form:

$H\left(z\right)=\frac{B_q\left(z\right)}{A_p\left(z\right)}=\frac{\underset{k=0}{\overset{q}{\mathrm{\Σ}}}{b}_q\left(k\right)z^{-k}}{1+\underset{k=1}{\overset{p}{\mathrm{\Σ}}}{a}_p\left(k\right)z^{-k}}$ Equation (1)

In some embodiments, the invention provides for will the Iodine contrast medium of adjustment contrast agent agreement be allowed to be administered to individuals patients or to be administered to the example of region of interest to be scanned.For determining that the method for patient's transfer function that contrast agent applies between position and interested enhancing position comprises model and relies on method and Model Independent method.Two kinds of methods or scheme are all system identification forms, and wherein this system comprises medicine and patient's (common factor comprises such as scan setting, region of interest and pathology).The result of two kinds of methods produces the estimation to contrast agent dynamics characteristic.The knowledge of system dynamic characteristic can such as be used in optimization step, determine to maximize signal to noise ratio (SNR) in this optimization step and minimize the infusion protocol of iodine load/dosage to patient (constraint of given injecting systems, Peak Flow Rate when comprising the specification of such as positive flow velocity, given contrast agent viscosity and attached conduit and radiography dosage).

A. the identification of existing model dependence

The exponent number that the people such as Bae devise contrast agent propagation reduces (or mixing) PK model.K.T.Bae, " Aortic and hepatic contrast medium enhancement at CT.Part I.Prediction with a computer model (carrying out the enhancing of the pre-timing aorta of CT part i and liver contrast agent with computer model) " of the people such as J.P.Heiken, Radiology, 207 (3) volume 647-55 pages, 1998, and K.T.Bae, " Multiphasic injection method for uniform prolonged vascular enhancement at CT angiography:pharmacokinetic analysis and experimental porcine model (evenly the extending the heterogeneous injecting method that vascular strengthens in CT angiography: drug metabolism dynamic Feature Analysis and experiment pig model) " of the people such as H.Q.Tran, Radiology, 216 (3) volume 872-80 pages, 2000, these documents are incorporated herein by reference.Modeling method in the document finds that the systemic physiological metabolic model of teaching in the people such as above-mentioned Bae, Heiken document of 1998 is too large and comprises too much the unknown, thus can not carry out feasible calculating based on each patient.Therefore Bae and colleague thereof are by the major part of single chamber approximating anatomy, and because first time is interested by strengthening dynamic characteristic, so removal blood capillary transfer chamber.The exponent number minimizing model of gained is shown in Figure 4.In the diagram, V is the liquid capacity of corresponding " chamber ", and C is the prediction concentrations in each " chamber ", and Q is the volume flow rate of the blood spreading all over health.Q and V estimates according to anatomical data.

Suppose that time course is continuous, will the single order coupled differential system equations of this model be described.

$V_v\frac{{\mathrm{dC}}_v\left(t\right)}{\mathrm{dt}}=Q_c{C}_c\left(t\right)-Q_v{C}_v\left(t\right)$

$V_r\frac{{\mathrm{dC}}_r\left(t\right)}{\mathrm{dt}}=Q_v{C}_v\left(t\right)+Q_s{C}_s\left(t\right)-Q_r{C}_r\left(t\right)$

$V_p\frac{{\mathrm{dC}}_p\left(t\right)}{\mathrm{dt}}=Q_r{C}_r\left(t\right)-Q_r{C}_r\left(t\right)$

$V_L\frac{d{C}_L\left(t\right)}{\mathrm{dt}}=Q_p{C}_p\left(t\right)-Q_L{C}_L\left(t\right)$

$V_s\frac{{\mathrm{dC}}_s\left(t\right)}{\mathrm{dt}}=Q_L{C}_L\left(t\right)-Q_L{C}_s\left(t\right)$ Equation (2)

When Differential Systems is changed into state space form, the order of the state matrix (A) (see equation (3)) obtained is less than the exponent number of this system.This order deficiency shows that it is singular point when attempting Inverted matrix itself.If expect that the transfer function (see equation 4) producing this system arranges with (after discretization) for parameter estimation, limit or controls, then this singular point can become problem.This system must by discretization, because CT measurement is inherent sampling process and gained signal strengthens curve reflection discrete time process.

$\stackrel{\→}{B}=\left[\begin{array}{c}\frac{C_C}{V_C}\\ 0\\ 0\\ 0\\ 0\end{array}\right]\stackrel{\→}{C}=\left[\begin{array}{ccccc}0& 0& 0& 1& 0\end{array}\right]\stackrel{\→}{D}=\left[0\right]$ Equation (3)

$\hat{G}\left(s\right)=\stackrel{\→}{C}(s\stackrel{\→}{I}-\stackrel{\→}{A})\stackrel{\→}{B}+\stackrel{\→}{D}$ Equation (4)

Another problem that exponent number reduces Bae model is that it can not catch recirculation dynamic characteristic to high fidelity.Even if applied by CT angiography (CTA) interested in dynamic characteristic in first time for us, the contrast agent recirculation peak value caught because being caused by systemic circulation and cardiac muscle is also useful.As indicated in the difference between Fig. 5 and Fig. 2 C, the difference demonstrated in system time constant and recirculation dynamic characteristic is compared to the output of Bae system and empirical data.

In order to overcome the intrinsic mathematical difficulties of Bae modeling method, disclosed in Wada and Ward model adjustment and come model be developed in the present invention.See " The hybrid model:a new pharmacokinetic model for computer-controlled infusion pumps (mixed model: the novel drugs metabolism dynamic performance model controlling infusion pump for computer) " of D.R.Wada and D.S.Ward, IEEE Trans Biomed Eng, 41 (2) volume 134-42 pages, 1994, the document was incorporated herein by reference.The model of Wada and Ward is formed as describing and controls alfentanil (strong analgesic) propagation in patient body (with reference to Fig. 6).Wada and Ward model allows to be transformed into discrete time-domain.The method is by conjunction with the delay (such as, contrast agent transmit in the lung pulse pipe time) in external transmission delay aids drug dynamic characteristic; With wherein by adding further chambers, to introduce the method for the people such as the Bae of delayed phase contrary.Each chamber in the model of Fig. 6 is formulistic by the input and output applying mass balance at this chamber.External medicament is incorporated into this system by inculcating input.The mass flow inculcated is added to right ventricle.Universal qualities equilibrium equation for a chamber is as follows:

$c_{\mathrm{out}}=\frac{1}{V_B}{x}_B$ Equation (5)

Wherein subscript B refers to blood room and T refers to the tissue compartment of 2 Room organ model.For the single chamber of such as blood vessel and ventricle, equation (5) reduces to an equation, because speed rates item k _tBand k _bTequal zero.K item (or speed rates item or coefficient) represents that material passes the diffusion of capillary-pipe film.Cl item represents the gap of the material from chamber.In the situation of the x-ray contrast agent drained by the glomerular filtration in kidney, Cl item is associated with the room of analogue kidney.Q _inand Q _outrepresent that blood enters or discharge the volume flow rate of this room, and C _xit is the variable of material concentration in interested room.This variable is main interest place, because the radiography in the concentration of contrast agent and vascular or organ strengthens linear correlation.

In one embodiment, algorithm structure of the present invention supposes that the blood volume of Fig. 6 middle chamber is similar to by such as making the tracing table that BMI, sex and age and center blood volume are relevant with cardiac output with cardiac output.Be difficult to pre-estimate the k parameter represented outside contrast agent diffusion to Ink vessel transfusing room.When attempting the contrast agent propagation representing CTA imaging applications, main " first time passes through " dynamic characteristic considering contrast agent, wherein pulmonary system rate transition parameter is most interested.Therefore, in figure 6, only 2 chamber units are for pulmonary circulation.Such as can from the parameter estimation based on public data, computer simulation and/or team innovation data (population kinetic data).The object of this algorithm estimates lung diffusion parameter by the experience of individual is strengthened curve fitting to model.The matching of k parameter can use the Simplex technology of such as Nelder-Mead algorithm or system identification technique (such as Steiglitz-McBride method) to carry out.Once cognitive phase completes, iterative numerical optimizing process can such as be used for determining to input agreement, this input agreement maximizes the iodine load (constraint of given injecting systems-such as, negative flow velocity, Peak Flow Rate are such as 6ml/s etc.) that image SNR minimizes patient simultaneously.

In Fig. 6, non-cardiopulmonary circuit assemblies is merged in systemic blood system (systemic blood block) by the simplification of model.This is reduced at first time of considering for CTA contrast agent application by being rational during dynamic characteristic, carries out in the second contrast agent application afterwards because scanning obtains.

Fig. 7 illustrates that the image of an embodiment of exponent number of the present invention reduction model represents.The mass transport relation of each chamber is represented by equation (5).In the figure 7, Qco represents cardiac output.Model in Fig. 7 can be used for describing contrast agent and propagates in individuals patients's body.Suppose to be aware of the height of patient, body weight and sex, estimate total center blood volume (CBV) by following formula:

Equation (6)

" Circulatory Physiology:cardiac output and regulation (physiology of circulation: the heart exports and regulates) " of A.C.Guytan, Saunders, Philadelphia, 173 pages, ISBN:07216436004.In equation (6), height in units of inch body weight then in units of pound.Also similar equation can be used to estimate cardiac output, but this carry out without the need to carrying out " test " system with the contrast agent of injecting on a small quantity by hypothesis.The static estimation unlikely actual flow characteristic (such as due to pathology) describing the patient checked with high-order fidelity in essence of global loops parameter.

Because the transmutability that each patient is intrinsic in estimating, the blood volume and the cardiac output that strengthen characteristic and can be used for estimating with Fig. 7 model better patient are injected in test.The method reduces the number of known variables by parameter estimation.The produced enhancing curve calculating of using that the people such as Mahnken test bolus administration of contrast agent on a small quantity by point factorial is estimated from the cardiac output of the individuality carrying out MDCT inspection.A.H.Mahnken, " Determination of cardiac output with multislice spiral computed tomography:a validation study (the using multi-slice spiral computed tomography visualization determination cardiac output: confirm research) " of the people such as D.Henzler, Invest Radiol, 39 (8) volume 451-4 pages, 2004, and A.H.Mahnken, " Measurement of cardiac output from a test-bolus injection in multislice computed tomography (carrying out the cardiac output of self-test bolus infusion with the measurement of multi-disc computed tomography visualization) " of the people such as E.Klotz, Eur Radiol, 13 (11) volume 2498-504 pages, 2003, both are incorporated herein by reference.The method adopted by people such as Mahnken is more early proposed by people such as Garret." Measurement of cardiac output by cine computed tomography (the measuring cardiac output by film computed tomography visualization) " of the people such as J.S.Garrett, P.Lanzer, Am J Cardiol, 56 (10) volume 657-61 pages, 1985, the document was incorporated herein by reference.When indicator is introduced in blood circulation, the volume blood flow of the bright flow circuits of Stewart-Hamilton relation table calculates as follows:

$Q_{\mathrm{CO}}=\frac{M_I}{\underset{0}{\overset{\∞}{\∫}}c\left(t\right)\mathrm{dt}}$ Equation (7)

Wherein M _ibe the gross mass of the indicator (or tracer) be expelled in loop and c (t) is measuring of concentration of indicator.For x-ray contrast agent, these values are to the gross mass of the iodine of patient injection and the contrast concentration in units of mgI/ml.Known linear relation (~ 25HU/ (1mgI/ml) (document of the people (2003) such as people (1998) and Mahnken, Klotz such as reference example Bae, Heiken described above)) because Hounsfield Units (Hounsfield Units) between (CT decay number) and contrast agent haemoconcentration, can carry out integration to obtain the denominator term in equation (7) to the time attenuation curve from CT scanner.

" Bolus Dynamics:Theoretical and Experimental Aspects (the injecting dynamic characteristic: theoretical and experiment aspect) " of M.J.K.Blomley and P.Dawson, The Brit.J.ofRadiology, 70 volume 351-359 pages, 1997, propose and allow to use following relation to estimate the geometrical point of the haemoconcentration from injection position to measuring position:

equation (8)

Wherein " quality _i" or M _iit is the quality of the iodine to patient injection." peak value enhancing " is the peak value (arbitrary unit, but be HU for CT research unit) of intensity and " blood flow volume " is blood flow volume between injection position and record position.In order to make unit keep correct, must by HU Conversion of measurement unit be become mgI/ml unit divided by conversion coefficient 25 [HU/ (mgI/ml)] by PeakEnh.The calculating of equation (8) allows the blood volume in drawing for estimate 7 in cardiopulmonary loop.Difference between equation (6) and equation (8) provides the blood values in systemic circulation chamber.Blood volume in cardiopulmonary loop (heart, lung and periphery injection room) can determine to scale based on anatomical data or estimate as follows.

Be different from axial CT dynamic scan Feischmann and the Hittmair method using the time of being recorded by the ROI being placed in descending aorta to strengthen curve, in some embodiments of the present invention, the time strengthens curve and produces from descending aorta and pulmonary artery.These two kinds of curves can be used for producing by the circulation time in lung loop and/or the estimation of other cardiorespiratory parameters.In addition, the tissue (being such as back to Collateral vein and other tissue as described below) be associated with contrast agent peripheral injection can not affect the calculating to parameter estimation subsequently.

Fig. 8 illustrates and strengthens the typical shaft of the level of curve to CT image in the time in Fig. 9 of scanning to produce subsequently.Like this, Fig. 9 illustrates and the Dynamic CT after in saliferous and salt-free 20ml contrast agent bolus to 64 years old women's body of 239 pounds, time is strengthened curve.In Fig. 9, the first curve illustrates the enhancement value in pulmonary trunk, and the second curve illustrates the enhancement value from ascending aorta.Pulmonary trunk strengthens and approximate (like this, contrast agent is fully diluted from RV to PA) in right ventricle.Because contrast agent recirculation, the signal in Fig. 9 does not turn back to baseline, but the amount that skew is proportional with the remaining contrast agent in blood flow.It is excessive that recirculation contrast agent can cause cardiac output and blood volume in equation (8) to estimate.In order to consider recirculation contrast agent, the gal function that curve fitting becomes following form can be strengthened by the in the future self-test radiography of injecting:

$C\left(t\right)=k{(t-t_0)}^a{e}^{\frac{-(t-t_0)}{b}}$ Equation (9)

Wherein k, a and b are fitting parameters.This matching is undertaken by least square technology.Then gained function can be used for the parameter estimation deriving cardiorespiratory system.Parameter to be estimated is transmission coefficient k _bTand k _tB, V _rH, V _lH, V _lungτ is postponed with passing through.These parameters can be combined in (θ=[k in vector theta _bT, k _tB, V _rH, V _lH, V _lung, τ]).The enhancing distribution that ROI place records is y _pA(n) and y _dA(n).Y _dAn the enhancing at () place strengthens (y in pulmonary artery _pA(n)) and the function-y of parameter vector θ _dA(θ, y _pA(n), n).The object of parameter estimation is to producing the estimation representing the θ of patient data best.Because there is the source of many changes and noise in parameter estimation, so have reason to suppose that estimation difference is Gaussian (Gaussian).Therefore maximum likelihood estimates that (MLE) can be used for estimating with cost function (cost function) derived parameter

$V=\frac{1}{2{\mathrm{\σ}}^2}{\mathrm{\Σ}}_{i=1}^N{(y_{\mathrm{DA}}^{\mathrm{meas}}\left(n\right)-y_{\mathrm{DA}}(\widehat{\mathrm{\θ}},y_{\mathrm{PA}}\left(n\right),n))}^2$ Equation (10)

Wherein it is estimated parameter vector.The best estimate of parameter vector is defined as:

$\widehat{\mathrm{\θ}}=\arg \underset{\widehat{\mathrm{\θ}}}{\min }{\mathrm{\Σ}}_{i=1}^N{(y_{\mathrm{DA}}^{\mathrm{meas}}\left(n\right)-y_{\mathrm{DA}}(\widehat{\mathrm{\θ}},y_{\mathrm{PA}}\left(n\right),n)}^2$ Equation (11)

The variance of parameter estimation vector is:

$\mathrm{cov}\left(\widehat{\mathrm{\θ}}\right)=F^{-1}$ Equation (12)

Wherein F is Fei Sheer (Fisher) information matrix, and it has proportional with V axle and reflects potential probabilistic characteristic value of parameter estimation.

Minimizing of equation (11) is undertaken by Levenberg-Marquardt algorithm or other numerical optimization technique.Parameters obtained estimates that then can be used for the prediction produced in Fig. 7 model strengthens.In order to determine minimum contrast medium simultaneously maximum signal strengthen the input function of (simultaneously considering the constraint of injecting systems), numerical value constrained optimization can be carried out to determine the best contrast-medium injection in Fig. 5.

B. the identification of Model Independent

Usually, Model Independent algorithm mainly data-driven, and do not need the parameter model in advance of system as above.The Model Independent identification using nonparametric spectrum estimator and parameter model is below described.

nonparametric spectrum estimator

Figure 10 and 11 illustrates respectively to patient's impulse response function h _testn () applies the result of direct Fourier analysis (by 64 rank fast fourier transform (FFT)) and Welch periodogram.The periodogram of Figure 10 of rectangular window is wherein used to disclose limit close to direct current.And, there is dominant pole in 0.22,0.16,0.12 and 0.03Hz.In Fig. 10, real data vector only has 45 points.It is 64 points that two kinds of shown signals are filled with zero-one, and another is 128 points.128 zero paddings are used not have a significant improvement resolution.As mentioned above, in this curve, there is 4 " limit ".Larger DC component can cover another limit near 0Hz.Whether not clear projection near 0.075Hz is limit.The time delay of signal main peak is the reason at an important physiologic information-analogue model zero point.

Because Welch periodogram is tending towards having better resolution, has better variance characteristic and usually have usually than periodogram better spectral leaks (leakage) characteristic, so made various Welch periodogram estimation device, to be used for determining whether to occur peak value near 0.075Hz and to be used for improving the resolution (peak value) of other limit.Three results shown in Figure 11 are as follows: (i) data vector has 50% overlapping Bartlett window (length=64); (ii) there is 50% overlapping Hanning window (length=64); And (iii) has 50% overlapping rectangular window (length=64).It is evident that and use the estimation device of Bartlett and Hanning window to disclose more details.This is not astonishing, because these windows decrease the spectral leaks that the real information hidden in signal can be made to obscure.The FFT on 256 rank can be used for the spectrum of Figure 11 to produce more data point and more level and smooth estimation.Find that the FFT being greater than 256 rank does not meaningfully improve resolution.

Bartlett window as shown in figure 11 and 50% overlapping Welch periodogram is used to disclose limit (and the comparatively small peak on the left of it) near 0.075Hz.In this estimation, there are 7 can distinguish limit.Bartlett window provides enough resolution and is not revealed as the less limit of decay (compared with estimating device with the Welch of use Hanning window).It is evident that, Welch method discloses the spectral composition of latent process more.

parameter model

Estimate suitable, consider drug metabolism dynamic characteristic and medicine dynamic characteristic and very useful in the suitable hypothesis of making based on the method for the reasonable model of data understanding.In addition, arma modeling can be used for setting up the best control chart transported of contrast agent.Figure 10 and 11 illustrates Feng Hegu, shows that signal model can comprise pole and zero more accurately.

Prony method is the method for the arma modeling coefficient of estimation finite data record.With reference to " Statistical Digital Signal Processing and Modeling (statistics signal processing and modeling) " of M.Hayes, New York, New York:Wiley and Sons, 1996,154-177 page, the document is incorporated herein by reference.Prony method hypothesis expects that the signal of simulation is the approximate impulse response-illustrate graphically in fig. 12 of this system.With reference to " Parametric Signal Modeling (the parameter signal modeling) " of J.H.McClellan, Advanced Topics in Signal Processing (Pentice-Hall, Englewood Cliffs, NJ, 1998) chapter 1, the document is incorporated herein by reference.This algorithm iteration solves best molecule and the denominator coefficients of the output minimizing in equation (13) the up-to-date production model being relevant to input signal (impulse response estimation) with regard to least square, a namely in equation (14) _pand b _q(using Levinson-Durbin recurrence).

${\mathrm{\ζ}}_{\mathrm{LS}}=\underset{n=0}{\overset{\∞}{\mathrm{\Σ}}}{\left|e^{\′}\left(n\right)\right|}^2$

Equation (13)

$\frac{\∂{\mathrm{\ζ}}_{\mathrm{LS}}}{\∂a_p\left(k\right)}=0$

$\frac{\∂{\mathrm{\ζ}}_{\mathrm{LS}}}{\∂b_q\left(k\right)}=0$

Equation (14)

The other method of the zero pole point modeling of unknown system is Steiglitz-McBride method, also referred to as iteration pre-filtering.The initial guess of the denominator coefficients in equation (1) is undertaken by using Prony method.Then the least square carried out between interested signal and the signal model pre-estimated minimizes, and repeats iteratively for several times (error is close to zero).Although do not find common convergence property for Steiglitz-McBride method, can notice that this technology restrains in 10 iteration.Reference example is as " Parametric Signal Modeling (the parameter signal modeling) " of J.H.McClellan, Advanced Topics in Signal Processing (Pentice-Hall, Englewood Cliffs, NJ, 1988) chapter 1.Other details of Steiglitz-McBride method can at " Statistical Digital Signal Processing and Modeling (statistics signal processing and modeling) " of Hayes, New York, New York:Wiley and Sons, find in 154-177 pages in 1996, the document is incorporated herein by reference.

Figure 13 illustrates the curve that Steiglitz-McBride method (running 10 iteration) spectral model for different rank compared with primary signal is estimated.Mean square deviation (MSE) between two kinds of signals drops to less than 25 for 10 rank or more, and the arma modeling showing to have in equation (6) 10 or more fully represents basic dynamic characteristic.Figure 14 illustrates compared with the impulse response data of Fig. 2 C for h _testthe time series of the Steiglitz-McBride estimation of the exponent number change of (n).It is evident that according to above result, the dynamic characteristic that Iodine contrast medium is injected describes by spectral analysis technique and uses ARMA signal modeling method to simulate.

Again, one or more physiological models as above (or any one model such as proposed by people such as Bae, or PHYSBE (can buy from the MathWorks company of MassachusettsNatick and such as combine the human recycle system's classical model can discussed at the Internet www.mathworks.com/products/demos/simulink/physbe from the SIMULIK product of MathWorks company purchase)) can be the mathematical model used in the present invention.Such as, one or more known outside patient's variable (such as height, body weight or blood pressure) can be input in model before beginning one's study, to provide the initial estimation therefore initial estimation of patient's impulse response also being provided to the response to imaging injection.This initial estimation can such as identifying or improving the structure for the parameter model or nonparametric model describing response curve before carrying out test injection.

In another embodiment, the flow path of such as arm vein or multiple fragments of patient's vascular can use can from MathWorks company buy the ANSYS bought from the ANSYS company of PennsylvaniaCanonsburg, the ALGOR bought from the ALGOR company of PennsylvaniaPittsburgh or the simulation of other application program.

Patient optionally carries out test injection, and record can be used for deriving the patient's impulse response recorded to the response of test injection.Then physiological models can be adjusted to and make patient's impulse response of prediction closer mate the patient's impulse response recorded.The method is more be sure of: be customized to picture injection and will produce the enhanced level process of expectation with the passing of time.There is provided known external variable to contribute to the adjustment of model to be constrained in patient, improve the matching of this model thus.

The improvement of the test injection information from the patient's pulse determined by model of the present invention is estimated enough, makes not need further model to change during imaging injection, and imaging is injected at from injection to terminating to carry out when calculating.Or except test injection change, or replace test injection, during imaging injection, other patient data that the measurement from one or more region of interest strengthens data and/or such as heart rate is used in the prediction changing model during imaging is injected.This change is specified to the change in picture infusion protocol/parameter while being therefore used in given injection, strengthen process with the expectation realizing better passing in time in one or more region of interest.

The patient parameter of such as heart rate or breathing rate such as can change because of patient's anxiety or uncomfortable increase in scanning process.Therefore, fetch heart rate or out of Memory from scanner or separate monitor and use this information determine to scale injection will advantageously.

In some medical procedures, need, in two or more body parts, there is different enhanced level.An example is coronary visualisation, wherein be desirably in coronary artery and there is high-contrast and simultaneously there is in right ventricle medium-contrast (and reaching possible contrast in left ventricle), make ventricular wall motion and arterial lumens diameter can flow to right ventricle and visual but without pseudomorphism in single pass because of concentrated contrast agent.In addition, to in the hepatic scan of cancer metastasis, can expect that the blood level of contrast agent increases steadily but not is fixed on certain level, thus can see that blood vessel increases (hyper vascular) or minimizing tumor has the contrast agent level strengthening or weaken compared with normal liver tissue.The slope expected can be selected according to the cancer types of expection.In addition, for some liver imaging research, expect that the arterial blood level of contrast agent reduced in the hepatic portal later stage (later portal hepatic phase).This result realizes by carrying out of short duration imaging injection at present.But due to the difference of patient, radiography enhanced level and contrast agent can not be controlled the time of advent very well, and produce non-optimal image.

In order to realize the enhanced level of the fixing of expectation or time to time change at multiple region of interest, if in time enough separately, various analytical method can be used.But if expect this system of enhanced level Over-constrained, then the analytical method of such as Fourier transform or other parts are mentioned herein method may run into difficulty.In these cases, can expect to be used in known optimizer or solver in mathematics or computer realm.An example investigates prediction to strengthen and the mean square deviation expecting to strengthen on the selected period.The method has the advantage that enhanced level only needs to define in imaging time, and some analytical methods need enhanced level to define on the whole research period, comprise rising and falling time.

A spendable solver example is that general roughness gradient (GRG2) nonlinear optimization developed by the Allan Waren of the Leon Lasdon of the University of Texas of Jane Austen and Cleveland state university is encoded.It is available in Microsoft Excel and MATLAB.Also able to programme or buy the solver of other Languages.

Solver also can be provided with other constraint of the maximum rate of change of such as minimum flow velocity, Peak Flow Rate, a limited number of flow velocity flex point, a limited number of stable flow velocity or flow velocity.These constraints can limit from injector operations, patient safety limits or other actual or convenient restriction is derived.Then, determine in other constraint, meet the injection distribution expecting contrast agent level best.Solver is particularly useful for deconvoluting and convolution method in time domain.It can be used for such as finding patient's impulse response according to test injection, then uses this patient's impulse response and expects that enhanced level finds optimal injection to distribute.

Device described herein, method and system are easy to realize with computer.The data be input in the model of device of the present invention, method and system can from such as operator, hospital information system, scanner or imaging device, syringe and/or one or more monitoring arrangement.Data optionally before (i) test injection, during (ii) test injection and/or afterwards and/or (iii) during imaging injection and/or provide afterwards.Data can automatically be supplied to computer or be inputted by operator.The example exported comprises for imaging device or scanner with for the timing of contrast medium injector or operating parameter.These export optionally automatically to communicate between related device, automated communications confirmed by operator or change, or read by operator, confirmation or change and transmit to be input to appropriate device.Automated communications path can relate to any one of multiple customization or industry standard communication protocols.Manual path can comprise for accurately and/or record keep for the purpose of print injection or scanner parameter or agreement.The U.S. Patent No. 6,397 of giving assignee of the present invention is such as being awarded in communication between syringe, scanner and/or miscellaneous equipment, and describe in 098, this patent is incorporated herein by reference.Being applicable to an example of protocols of the present invention is the controller automatic network (CAN) such as described in ISO11898.The imaging system of imaging device (scanner) also communicates by operator with syringe.Like this, imaging device such as also can export injection distribution by Modling model on the display apparatus.Then injection distribution can be input in syringe by operator.

The algorithm and model of device of the present invention, system and method can such as implemented with in one or more computer implemented one or more computer program.Be applicable to exemplary computer language of the present invention or environment includes but not limited to the ACSL that Excel, Visual Basic, Matlab, C++ and the Aegis Technologies Group by Alabama Huntsville make.The computer realizing program of the present invention can be an add ons of a part for such as imaging device, a part for syringe or imaging system.Or multiple operation can be distributed between the computer that associates with two or more device.

Figure 15 A illustrates arm vein and leads to total figure of vein of heart.Figure 15 B illustrates some changes of arm vein anatomical structure.It is evident that, the path of one or more can be adopted to arrive heart at the liquid of the hemostasis of bifurcated.These paths can have different length and resistance, thus cause the different transmission times.Figure 16 A and 17A illustrates that the pulmonary artery (PA) of at least latter two different patient of the saline injection of 60ml under the test injection of 20ml under 4.5ml/s and 4.5ml/s subsequently strengthens.In Figure 16 A, the enhancing curve of patient 1 illustrates single peak, shows that contrast agent adopts individual paths or has multiple paths of similar transmission time.In Figure 17 A, the enhancing curve of patient 2 illustrates because contrast agent adopts the path and two resolved peaks producing with the different transmission time.In order to this is described, an embodiment (such as shown in Figure 7) of chamber model can comprise two parallel chambeies with not co-content and different in flow rate, to allow the curve modeling more accurately to Figure 17 A.Figure 16 B and 17B illustrates that the ascending aorta (AA) for same two patients strengthens curve.Patient's 2 is bimodal level and smooth by transmission in lung.If therefore only carry out AA measurement, the ability of correct simulation arm stream will be lost.

As mentioned above, many devices described herein, system and/or method use test injection to determine about the information of patient to the response of drug injection.For many models, test injection can have Arbitrary distribution, as long as this is distributed as known.For some medicines, flow velocity and the enough low so that test injection of amount can not disturbances or change the flow velocity of blood in vein.This represents that this medicine decelerates to Hemodynamic environment very soon, and is transported to central vasculature with blood flow rate.Use the x-ray contrast agent being used for imaging, flow velocity is with amount usual enough high (some milliliters per second) so that significant disturbance occurs.In this case, once the injection of contrast agent is slack-off or stop, the contrast agent still flowed in peripheral vasculature will consume its momentum and slow down, because it is now only subject to the driving of blood and deceleration contrast agent flow (if existence).In order to overcome these, the test injection of contrast agent can be carried out under constant flow rate, and the non-radiography injecting the such as saline of some seconds subsequently with same flow velocity strengthens liquid, so that contrast agent is displaced peripheral vasculature.Figure 18 A and 18B illustrates the example that flow reduces and total effective dose reduces occurred when not comprising flushing liquor for two patients.Injection programming stream mainly realizes with normal saline washing, because all contrast agent are discharged from peripheral vasculature by normal saline washing.With regard to patient A, for the injection without normal saline washing, arrive 13.5 milliliters in the 17ml that the effective dose of central vasculature is only injected.And be injected at and only start after 3 seconds to reduce, and extend to 13 seconds.With regard to patient B, for the injection without normal saline washing, effective dose is the 15ml in 17ml, but never reaches actual flow velocity.Therefore, best test is injected and can be defined by having injecting of constant volumetric flow rate, the concentration of its Chinese medicine or active component as programme change (such as rectangular pulse, gaussian-shape or random waveform), the concentration comprising active component is the flow of after zero some seconds.

In some cases, to carry out twice or test injection is repeatedly useful, once another time flushing liquid without flushing liquid.Test injection more than use once provides the instruction about the distal venous drain providing information when determining optimal imaging injection parameters.

Some patients have the central venous catheter or PICC line (centre pipe that periphery is inserted) that can inject, thus contrast agent need not flow in branch terminal vascular system.In some cases, this simplifies modeling to a certain extent and accelerates contrast agent transmission.In these cases, because its behavior is known, conduit can be simulated clearly.All of the embodiments of the present invention is suitable for operating at various injection position, and some of them can allow carry out test injection a position and carry out imaging injection at a diverse location.

During imaging injection, the flow velocity (milli Grams Per Second) of contrast agent molecule affects by three kinds of modes.The first, the volume flow rate (ml/S) of constant density contrast agent can change.The second, the volume flow rate in units of milliliter is per second keep constant while can change the concentration (amount of contrast agent molecule, mg/ml) of active component in contrast agent.3rd, volume flow rate and concentration both alterable.The first selection only uses a kind of liquid to realize by simple injecting systems.This has above-mentioned difficulty: the contrast agent in patient's arm fails to be convened for lack of a quorum along with entering the reduction of contrast agent flow or increase and reducing or increase.Under steady flow, the dilution of contrast agent preferably changes flow velocity or speed, because this maintains " driving stream " of or having caught with model or algorithm.This performance is to such as U.S. Patent No. 5,840,026,6,385,483 and 5,494, and the cartridge syringe in 036 is possible, and these patents are incorporated herein by reference.Can buy from the Medrad company of Pittsburgh of Pennsylvania syringe can such as realizing the high pressure high flow rate of saline and contrast agent.The third selection in practice can be better, because it allows there is the high flow velocities higher than common 3-5ml/S when needed, and such as 6-10ml/S.This third select also permissible velocity of flow to be reduced to more appropriate and safer a little flow velocity.But if contrast agent flow needs below the lower limit of such as 3ml/S, then this result realizes the constant drive of with maintenance, contrast agent being discharged peripheral venous circulation by dilution.

By having the initial high flow velocities reduced in time, overflowing unheeded probability and being reduced.Nurse can such as palpation injection position within former second.If overflowed for higher initial flow, then it can not occur under comparatively low flow velocity.Or overflow detector is more responsive to the faster signal produced because of the spilling under high flow capacity, and therefore more likely stop injection when overflowing and occurring.

One Additional embodiments can utilize the concentric tubes comprising inner chamber or center cavity and exocoel, such as combine and award the serial number No.10/821 giving assignee of the present invention, disclosed in Figure 11 A-11C of the U.S. Patent application of 210 (open U.S. Patent application No.2004/025452), this application is incorporated herein by reference.Two chambeies of this conduit can be arranged in stream from exocoel substantially around the stream from inner chamber.Wherein in the situation of image-forming contrast medium than blood or saline thickness more, image-forming contrast medium by center cavity conveying, can carry in exocoel by saline or non-reinforcing liquid.This contributes to reducing to promote contrast agent by peripheral vein and the pressure entered needed for central venous system.As described herein, contrast agent and non-contrast can regulate as required, to keep the constant volume stream in the vascular of periphery when contrast agent molecule transfer rate changes.This concentric tubes also can be passed through Rule adapter (luer adapter) valuably and be connected to PICC line or central guiding pipeline, to provide contrast agent flow in salt current, this reduces the pressure drop in PICC or central guiding pipeline.

If test injection comprises the injection of non-reinforcing liquid at the end of it, then syringe " must be remembered " and solve any adapter pipeline used and centre pipe will be filled with non-radiography and strengthen liquid.Similarly, when first preparing syringe for patient, the adapter pipeline between syringe and patient can be filled with contrast agent or saline, and this syringe needs to solve this.Otherwise arrive occurring the unexpected morning of contrast agent or arrival in evening, and the impulse response obtained or model applicability are by inaccurate.Do not solve these to make in the algorithm of adaptive response and being especially a problem during imaging injection.

Desired image intensity data reasonably accurately and can repeat.But many factors can affect the absolute accuracy of the process of the 2D image of image reconstruction algorithm and generation Hounsfield Units.Such as KV (kilovoltage on electron tube), FOV (visual field), bench height and restructing algorithm are the known variables affecting precision in CT.Therefore, expect to use compatible restructing algorithm and other variables set.But, use modern scanners, there is many automatic dosage minimizing methods that imaging variables is changed in scanning process.In addition, use ECG gating (gating), restructing algorithm can change piecewise.Once overcame this variable method to adapt according to quantitive CT (QCT).Bone mineral standard or calibrating patterns (phantom) are placed in be had in the scanner of patient, and as described in the brochure SL172E 7/01 (copyright 2001) of GE Lunar company, its disclosure is incorporated herein by reference.The method allows to convert the Hounsfield Units of reconstruct to absolute Hounsfield Units.Second method uses the fat of patient, skeleton and/or muscular tissue but not interested outer portion in calibration criterion.The method is only effective in some cases, because these are organized was only constant Hounsfield Units before contrast agent arrives this tissue.Such as, it may be lung in interested imageable target or position, the heart or large vascular effective when using the organizing of such as spinal column and esophagus.Tissue of patient is used for calibrate this use to test injection or when imaging injection starts than more practical to the middle and late stage of imaging injection.

QCT standard such as can be based upon in liner on patient support platform, platform and/or scanner gantry or become their part.It needs not to be concrete model structure, such as, be generally used for building the aluminum of patient sick bed, carbon tube/epoxy resin and foamed materials just enough, as long as their as one man attenuate X-rays.In these situations always at image ad-hoc location of standard, restructing algorithm can automatic access standard value correct patient image.QCT standard also can not appear in the region in patient image.

Even if precision improvement as described above, because the statistics of X-ray absorbs, in Hounsfield Units calibration, still there is noise.Region of interest large is as far as possible used to reduce or minimize this effect.High frequency change or noise are tending towards reaching average.Also more complicated area computation method can be applied.Other method is by the measured value of curve fitting in the Hounsfield Units of relative time.Normally used curve is gal variable curve.Also polynomial curve can be used.This is level and smooth noise time between actual measurement provide the estimation to strengthening.The experiment curv of Figure 16 A-17B represents that the meansigma methods of ROI in vascular (region of interest) adds a standard deviation and deducts a standard deviation.

Other method for improvement of accuracy is in continuous print region of interest by this models fitting.In a CT sheet, likely such as, to some different parts imagings, left ventricle, lung tissue, right ventricle and descending aorta.If strengthened in Hounsfield Units by this models fitting when contrast agent flows through different parts, then can realize the more accurate matching as above described in composition graphs 16A-17B.This can be applied to based on physiological model potentially most effectively, because this model is normally segregative and provide the method strengthened the prediction at the distinguishable position of interested physiology.

The some embodiments of the present invention as herein described relate generally to the desired image contrast agent patient response realizing Hounsfield Units.The example of desired image contrast agent patient response is as the relative constancy sometimes desired by blood vessel imaging or smooth enhancing, as to the blood contrast agent level of the useful continuous rising of Metastatic Liver cancer detection and the contrast agent level that allows the rising of Portal venous phase (portal venous phase) liver imaging then to decline.But, the present invention be convenient to realize doctor diagnose needed for usually strengthen distribution arbitrarily, such as functional imaging or perfusion map.

In addition, strengthen and occur in the concrete time period.Because carry out the method analyzed, time zero is arbitrary.In one embodiment, expect to strengthen be limited at from 0 enough away from moment, after being wherein injected at for 0 moment the long duration just need start.Such as, enhancing may be selected to and reached aspiration level at 100 seconds.Then, once optimize patient or build model, then this model is by the time started of prediction imaging injection, such as 70 seconds.In use, injection starts and expects that strengthening the difference between starting is used to determine delayed sweep, is 30 seconds in this example.Once everything is ready, syringe just starts (Part III from syringe, scanner or equipment), and scanner starts to perform scanning by programming series (again, being triggered by the Part III of syringe, scanner or equipment) after 30 seconds.

Many embodiments of the present invention all describe (see such as Figure 19 A and 19B) relatively with being injected into the adjustment to patient model, the change carried out of picture based on test injection imaging or image or upgrading.In addition, common patient model or many patient model can regulate based on the result of one or more patient, change or upgrade, for other patient.If the embodiment realized does not use the real-time adjustment of test injection or injection parameters, then this will be especially applicable.In this case, the patient's design parameter used is all known before beginning imaging, and can comprise such as morbid state, height, body weight, approximate cardiac output (normal, shortage, deficiency) and other usually known parameter.

Exemplary embodiment of the present invention various embodiments that are main and CT imaging describe relatively.But, those skilled in the art should understand that: device of the present invention, system and method are easy to allow to improve optimal dose in other imaging devices all and method for (through change or without the need to change), comprise such as nuclear magnetic resonance, ultra sonic imaging, X-actinoscopy X, positron emission computerized tomography and various photoimaging equipment.In order to use in except other imaging process of CT imaging, to any change of the present invention all in the technical scope of those skilled in the art.

Various medicine can be benefited from application of the present invention.This includes but not limited to ion, nonionic, dimer and blood pond contrast agent.Being also included within interior is physiologically active medicine, especially have short-decayed those.Two or more difference or similar model can use in same diagnostic procedure together with two or more different pharmaceutical, and such as CT, MR or ultrasonic image-forming contrast medium combine with the dobutamine (dobutamine) for cardiac pressure imaging.

Imaging or image enhaucament level are the examples exported from " sensor ".Be applicable to other sensor of the present invention in U.S. Patent No. 5,840,026 and 6,385, list in 483, these two patents are incorporated herein by reference.Other sensor relates to other physiological parameter, such as the blood level of such as chemotherapeutical drug blood level or the such as pharmaceutical product of blood glucose or blood clot dissolving molecule.Be applicable to other sensor example of the present invention include but not limited to: EEG (electroencephalogram), muscle response sensor or specific nerve tract activity level sensor.

In some imaging devices, object is in order to radiography enhanced level is kept the time longer than several seconds.An example is ultra sonic imaging.These contrast agent medicines are conventional blood pool agents, represent that they are detained in the blood vessel and do not diffuse into outside blood vessel or intracellular space.Therefore, the mathematical model of deriving is different in detail for the mathematical model in CT from derivation, but device disclosed herein, system and method are still useful and applicable.

When developing a model to patient, it is recorded and preserves, and makes it become the basis of model subsequently.Sometimes this can eliminate test subsequently, such as, eliminate the needs to test injection.In addition, it can increase the accuracy of model.In addition, record and preserve test scan measured value or image contribute to after the model of scanning set up.

Hypothetical model/patient's transfer function is identified (form such as with equation (1)), can attempt the input signal solved producing desired output (the expectation radiography enhanced level namely in the region of anatomy interested).Suppose a linear time-invariant system (being also cause and effect or stable), the Input output Relationship of discrete time (or sampling) system is

equation (15)

Wherein operator it is inverse discrete Fourier transformer inverse-discrete.Because this system is assumed to linearly, the time is constant and cause and effect, so the item in the summation of right side can be exchanged, Input output Relationship also can be write as:

$y\left(n\right)=\underset{m=0}{\overset{n}{\mathrm{\Σ}}}h\left(\hat{m}\right)x(n-m)$ Equation (16)

H (z) item is calculated by the modeling in advance of patient/drug system as above, by calculating the system identification technique of the data operation collected in the brief inquiring process of the system at a small amount of drug injection or use the combination of two kinds of methods to calculate.Suppose that noiseless is measured, equation (15) can be rewritten as linear algebraic equation:

$\stackrel{\‾}{y}=\stackrel{\‾}{\stackrel{\‾}{H}}\·\stackrel{\→}{x}$ Equation (17)

Wherein the column vector of long M x 1, it is the column vector describing to length the value being the input of the system responses of the model of cognition of N (supposing M=N).Matrix be hereby (Toeplitz) matrix in lower triangle Tobe, there is following structure:

equation (18)

Given desired output ( vector) and be expressed as the Toeplitz matrix of equation (18) form for vector a kind of method of (input) solving equation (17) finds to make the vector x of cost function minimization (being called Linear least squares minimization problem):

$J\left(\stackrel{\‾}{x}\right)=\frac{1}{2}{\left|\right|\stackrel{\‾}{y}-\stackrel{\‾}{\stackrel{\‾}{H}}\stackrel{\‾}{x}\left|\right|}^2=\frac{1}{2}{(\stackrel{\‾}{y}-\stackrel{\‾}{\stackrel{\‾}{H}}\stackrel{\‾}{x})}^T(\stackrel{\‾}{y}-\stackrel{\‾}{\stackrel{\‾}{H}}\stackrel{\‾}{x})$ Equation (19)

Cost function in equation (19) realizes its global minimization when its gradient equals zero.A known result is that vector x can solve into:

$\stackrel{\‾}{x}={\left({\stackrel{\‾}{\stackrel{\‾}{H}}}^T\stackrel{\‾}{\stackrel{\‾}{H}}\right)}^{-1}{\stackrel{\‾}{\stackrel{\‾}{H}}}^T\stackrel{\‾}{y}={\stackrel{\‾}{\stackrel{\‾}{H}}}^{+}\stackrel{\‾}{y}$ Equation (20)

Wherein it is a mole-Peng Nuosi (Moore-Penrose) pseudo inverse matrix (pseudo-inverse).Equation (20) is the general solution when the row rank of H exceedes column rank to overdetermination situation.When row rank equal with column rank (and being therefore square matrix) time, the solution of equation (19) is:

$\stackrel{\‾}{x}={\stackrel{\‾}{\stackrel{\‾}{H}}}^{-1}\stackrel{\‾}{y}$ Equation (21)

Wherein the inverse matrix of H is by many technique computes (Gauss with pivoting cuts unit, singular value decomposition etc.).Obviously, if the impulse response of model matrix irreversible (condition is deteriorated due to singular point or due to noise), then can not determine the reliable input signal that can realize desired output.This is gratifying adjustment intuitively, because if the system of equation 1 is irreversible, linear, the time is constant and causal condition is no longer subject to forcing (or real processes can not be approximate by these hypothesis).It should be noted that significant noise and/or model uncertainty can cause the relatively large deviation in the numerical solution of equation (21).

Also can using is tied deconvolutes solution (rule that formula changes into the optimization problem that is tied is deconvoluted), and wherein minimized expression-form (DeNicolao is G.1997) is as follows:

$\underset{u\≥0}{\min }{(y-\mathrm{Hu})}^T{B}^{-1}(y-\mathrm{Hu})+{\mathrm{\γu}}^T{F}^T\mathrm{Fu}$ Equation (21)

Equation (21) does not have enclosed (closed-form) and separates, so it must be solved by iterative technique.H matrix be impulse ratio lower triangle Toeplitz matrix-it strengthens the value of curve from the time of test injection imaging by System identification or by insert simply and determines.Y item represents desired output and u controls.In equation 21, Section 2 represents noise corrupted and probabilistic means in solution measurement.Matrix F represents " forgeing " factor on its diagonal with noise covariance estimation.Equation (21) can such as practical weighted least-squares numerical optimization and/or the multiple target optimisation technique that is tied solve.

Although the present invention has been described in detail in conjunction with above-mentioned embodiment and/or example, should be appreciated that this details is illustrative and nonrestrictive, and those skilled in the art can carry out changing and not deviating from the present invention.Scope of the present invention by appended claims but not above description instruction.Drop on all changes in the implication of the equivalents of each claim and scope and change and be all included within the scope of it.

Claims (50)

1. radiography is strengthened the controller that liquid flows to patient by operating syringe system, comprising:

A at least one patient's transfer function that () is determined described patient based on the concrete data pin of patient, at least one patient's transfer function described represents the propagation of radiography enhance liquid in described patient and provides the expected time to strengthen to given input and exports, and the concrete data of described patient comprise the estimation of physiological parameters of patients, the estimation of patient anatomy parameter and by least one in the enhancing caused patient injection radiography enhance liquid; And

(b) processor, for using at least one patient's transfer function described to determine that injection process inputs, described injection process input strengthens output by generating the described expected time determined for described patient,

Wherein said controller is used for operating described injector system based on the injection process input determined by least one patient's transfer function described at least in part.

2. controller as claimed in claim 1, is characterized in that, described injection process input is determined by least one performance constraint or constraint of considering described injector system.

3. controller as claimed in claim 2, is characterized in that, at least one patient's transfer function described is determined by using the system identification model comprising the parameter relevant to the described physiological parameter of described patient.

4. controller as claimed in claim 3, it is characterized in that, described system identification model can discretization.

5. controller as claimed in claim 3, it is characterized in that, described controller is used for:

Use and initial patient's transfer function is formed to the estimation of at least one physiological parameter of described patient; And

At least one time based on the described injection of collecting during injecting or after injection strengthens to export revises described patient's transfer function.

6. controller as claimed in claim 5, is characterized in that, at least one physiological parameters of patients strengthens output measurement according at least one time described and obtains.

7. controller as claimed in claim 5, is characterized in that, the test injection that described injection is carried out before being diagnosing image process.

8. controller as claimed in claim 7, is characterized in that, the time enhancing output produced because of described test injection obtains at least two region of interest measurements.

9. controller as claimed in claim 8, is characterized in that, the described time strengthens the difference between exporting provides measuring of at least one physiological parameters of patients.

10. controller as claimed in claim 9, it is characterized in that, at least one physiological parameters of patients described is the parameter of cardiorespiratory system.

11. controllers as claimed in claim 10, is characterized in that, at least one physiological parameters of patients described is blood volume in cardiac output, position, speed rates item or by postponing.

12. controllers as claimed in claim 8, is characterized in that, the very first time strengthens to export measures and obtain in ascending aorta or descending aorta, and the second time strengthened output and measures in pulmonary trunk and obtain.

13. controllers as claimed in claim 9, is characterized in that, the very first time strengthens to export measures and obtain in ascending aorta or descending aorta, and the second time strengthened output and measures in pulmonary trunk and obtain.

14. controllers as claimed in claim 13, is characterized in that, at least one physiological parameters of patients described is blood volume in cardiac output, position, speed rates item or by postponing.

15. controllers as claimed in claim 2, is characterized in that, at least one patient's transfer function described is determined by following steps:

Collect the data corresponding with curve time response that the injection strengthening liquid because of described radiography produces; And

Determine at least one mathematical model describing described data.

16. controllers as claimed in claim 15, is characterized in that, at least one mathematical model described is not determined by the continuous of described data or discrete time Fourier conversion.

17. controllers as claimed in claim 15, it is characterized in that, at least one mathematical model described is parameter model.

18. controllers as claimed in claim 17, it is characterized in that, at least one mathematical model described is rolling average or auto regressive moving average.

19. controllers as claimed in claim 17, is characterized in that, described at least one mathematical model hypothesis is linear and the time is constant.

20. controllers as claimed in claim 15, it is characterized in that, at least one mathematical model described is the nonparametric model determined by spectrum estimating techniques.

21. controllers as claimed in claim 20, is characterized in that, described spectrum estimating techniques are Welch method, Bartlett method, Multiple Signal Classification (MUSIC) method or period map method.

22. controllers as claimed in claim 15, is characterized in that, collect during the test injection at least one times of described data before imaging injection.

23. controllers as claimed in claim 15, is characterized in that, the Data Update that at least one patient's transfer function described is collected during being used in imaging injection.

24. controllers as claimed in claim 3, is characterized in that, the Data Update that at least one patient's transfer function described is collected during being used in imaging injection.

25. controllers as claimed in claim 2, is characterized in that, at least one patient's transfer function described is determined based on injecting at least one times at least in part.

26. controllers as claimed in claim 25, is characterized in that, described injection is at least one times the test injection of carrying out before diagnosing image process.

27. controllers as claimed in claim 26, is characterized in that, described test injection comprises described radiography and strengthens liquid injection and non-contrast liquid infusion afterwards.

28. controllers as claimed in claim 27, is characterized in that, described non-contrast liquid is to inject with the volume flow rate that the described radiography before the described injection of described non-contrast liquid strengthens the flow velocity of liquid substantially the same.

29. controllers as claimed in claim 28, it is characterized in that, described non-contrast liquid is saline.

30. controllers as claimed in claim 26, is characterized in that, carry out once above test injection.

31. controllers as claimed in claim 30, is characterized in that, test injection only includes described radiography and strengthens liquid injection, and another test injection comprises injection and non-contrast liquid infusion subsequently that described radiography strengthens liquid.

32. controllers as claimed in claim 2, is characterized in that, described injection process input is by using analytic solutions or using numerical value constraint optimization to determine.

33. controllers as claimed in claim 32, it is characterized in that, described numerical value constraint optimization is weighted least-squares numerical optimization.

34. controllers as claimed in claim 32, is characterized in that, described injection process input is optimized to the quality making the described radiography flowing to described patient strengthen contrast medium in liquid and minimizes.

35. controllers as claimed in claim 34, it is characterized in that, described contrast medium is iodine, xenon or gadolinium.

36. controllers as claimed in claim 1, is characterized in that, described radiography strengthens liquid can take following form: be applicable to the contrast agent of CT imaging process, be applicable to the contrast agent of MRI imaging process or be applicable to the reinforcing agent of ultrasound imaging procedure.

37. controllers as claimed in claim 1, it is characterized in that, at least two patient's transfer functions are determined, and the input of described injection process is determined based on one of described patient's transfer function.

38. controllers as claimed in claim 37, it is characterized in that, first patient's transfer function is determined by using the system identification model comprising the parameter relevant to the described physiological parameter of described patient, and second patient's transfer function is by using by collecting with because injecting that the time produced strengthens the corresponding data of curve, the mathematical model determined is determined, described mathematical model describes described data.

39. controllers as claimed in claim 37, is characterized in that, to which in first patient's transfer function and second patient's transfer function provide given input and result export between optimal relevance make and determining.

The controller of 40. 1 kinds of operating syringes in medical imaging procedures, described controller is used for:

A () determines that the time that at least one mathematical model produces because of contrast-medium injection with prediction strengthens response;

B () determines that an infusion protocol strengthens response with the scheduled time by determining the constraint input of at least one mathematical model described to separate in approximate patient body; And

C () uses described infusion protocol to operate described syringe in described medical imaging procedures, to operate the image setting up region of interest in described contrast-medium injection to described patient body.

41. controllers as claimed in claim 40, are also included in imaging process and change described infusion protocol because of strengthening the feedback of response about the described time.

42. controllers as claimed in claim 40, are also included in described imaging process and change described infusion protocol because of the data about at least one physiological parameters of patients.

43. controllers as claimed in claim 40, is characterized in that, determining that described infusion protocol strengthens response with the approximate described scheduled time is by using numerical value solver or numerical optimization device.

44. controllers as claimed in claim 40, is characterized in that, at least one performance constraint that the constraint input of at least one mathematical model described is separated by described syringe retrains.

45. controllers as claimed in claim 40, is characterized in that, the constraint input of at least one mathematical model described is separated by retraining to patient safety or comfortable at least one relevant performance constraint.

46. controllers as claimed in claim 40, is characterized in that, described contrast-medium injection in a moment, and the second moment that the image scanning of described region of interest is being determined based at least one mathematical model described at least in part start.

47. controllers as claimed in claim 46, is characterized in that, described second moment is determined based on to the moment prediction reaching the predetermined enhanced level determined by least one mathematical model described.

48. controllers as claimed in claim 40, is characterized in that, at least one mathematical model described is patient's transfer function of the described patient based on the concrete data of patient, and at least one patient's transfer function described provides the time to strengthen to given input and exports.

49. controllers as claimed in claim 48, it is characterized in that, described patient's transfer function is determined by using the system identification model comprising the parameter relevant to the physiological parameter of described patient, or by using by collect and the time that the injection because of described patient produces strengthens the mathematics model of cognition that the corresponding data of curve determine and determines, wherein said mathematics model of cognition describes described data.

50. 1 kinds of imaging systems:

A () sets up the imager of the image of patient's region of interest;

B () is suitable for the syringe of injection of contrast medium; And

C () and described syringe efficient communication are to operate the controller of described syringe, described controller comprises:

I at least one patient's transfer function that () is determined described patient based on the concrete data of described patient, at least one patient's transfer function described provides the expected time to strengthen to given input and exports, and the concrete data of described patient comprise the estimation of physiological parameters of patients, the estimation of patient anatomy parameter and by least one in the enhancing caused patient injection contrast agent; And

(ii) processor, strengthens output to use at least one patient's transfer function to the described expected time and determines that injection process inputs.